git: unable to reduce .git folder size

Question

This is the size of my .git folder:

$ du -sh .git
321M    .git

There are a couple of huge files accidentally commited there in the past.

Through some bizarre magic I found the cuplrits and their blobs:

$ git rev-list --objects --all |   git cat-file --batch-check='%(objecttype) %(objectname) %(objectsize) %(rest)' |   sed -n 's/^blob //p' |   sort --numeric-sort --key=2 |   cut -c 1-12,41- |   $(command -v gnumfmt || echo numfmt) --field=2 --to=iec-i --suffix=B --padding=7 --round=nearest


....
5c48fadaee41  492MiB android/java_pid3568.hprof
...

And the author of these commits:

git log --all --full-history -- "android/java_pid3568.hprof"

commit cac160ab39b9c70f3d05f8194be8c3c0657161ad
Author: Incore <xxxx@mail.ru>
Date:   Thu Dec 10 18:34:34 2020 +0300

commit fc0cfc2f32ddd0f927169ad0514213f79795dd63
Author: Incore <xxxx@mail.ru>
Date:   Tue Dec 8 21:26:17 2020 +0300

After this, I tried to rewrite my history removing this file:

git filter-branch --force --index-filter   'git rm --cached --ignore-unmatch android/java_pid3568.hprof'   --prune-empty --tag-name-filter cat -- --all



Rewrite fc0cfc2f32ddd0f927169ad0514213f79795dd63 (19/376) (1 seconds passed, remaining 18 predicted)    rm 'android/java_pid3568.hprof'
Rewrite 540ced9097b7090bfd68279515178bf060db93ee (19/376) (1 seconds passed, remaining 18 predicted)    rm 'android/java_pid3568.hprof'

Now my git log shows there is no such file in history:

git log --all --full-history -- "android/java_pid3568.hprof"

... empty response

However, the size of my .git folder has not been reduced:

$ du -sh .git
321M    .git

why is that?

I'm not entirely sure how filter branch works, but if any commit which still contains those files is reachable, then the git repo won't reduce in size. — evolutionxbox, Feb 26 '21 at 14:43
if `git log` searching for those files shows empty response - this should mean that there aren't any commits containing this file, are there? — kurtgn, Feb 26 '21 at 15:06
Try using `git log` on every branch? local and remote branches — evolutionxbox, Feb 26 '21 at 15:07
thanks for suggestion, but nothing there https://prnt.sc/107ofsu. Locally I only have master — kurtgn, Feb 26 '21 at 15:42

Philippe · Accepted Answer · 2021-02-26T23:58:08.950

Commits containing the huge file are still handled by the reflog so they can't be garbage collected just by a git gc.

You could find some good commands in the answers to this question: How to remove unreferenced blobs from my git repo

like

git reflog expire --expire-unreachable=now --all
git gc --prune=now

or

git -c gc.reflogExpire=0 -c gc.reflogExpireUnreachable=0 -c gc.rerereresolved=0 -c gc.rerereunresolved=0 -c gc.pruneExpire=now gc

PS: oh, and it's always a good idea to use https://rtyley.github.io/bfg-repo-cleaner/ to do this kind of cleaning. Much easier, secure and quicker than filter-branch.

git: unable to reduce .git folder size

1 Answers1